Research Scientist (Physical AI)

About the role

Nomagic is building general‑purpose physical AI by leveraging massive deployment logs from robots operating in production. As a Research Scientist you will bridge cutting‑edge machine‑learning research and large‑scale robotic execution, developing foundational models that transform classical control stacks into AI‑driven agents.

Key responsibilities

• Design and implement transformer‑ and diffusion‑based model architectures for multimodal robotic data.
• Develop scalable data mixing, sampling strategies and training curricula using vision, action and state logs.
• Conduct rigorous ablations to study scaling laws, data quality, optimization dynamics and failure modes of large models.
• Collaborate with engineering teams to integrate and deploy models on real robot fleets.
• Iterate on pre‑training and post‑training pipelines to improve agent performance and economic impact.

Required profile

• PhD or equivalent experience in machine learning, robotics, or a related field.
• Strong research background with publications in deep learning, multimodal models, or robotics.
• Experience scaling models on large datasets and distributed training environments.
• Ability to translate research findings into production‑ready systems.

Required skills

• Python programming
• Deep learning frameworks (e.g., PyTorch, TensorFlow)
• Transformer and diffusion model architectures
• Multimodal data processing (vision, action, state)
• Large‑scale model training and distributed computing
• Robotics systems and control concepts
• Data engineering and pipeline development

What we offer

• Hands‑on work with real robots solving real problems daily
• Relocation package
• Flexible working hours
• English‑speaking environment